What's behind the decreasing college graduate unemployment advantage?
Between 2006 and 2025, labor force participation among non-college young people fell 2.6 percentage points. The rate for young college graduates also fell, but only by 0.8 percentage points.
In recent months, there has been a flurry of news reports about an unemployment crisis among recent college graduates in the United States. Many commentators suggest that recent graduates are struggling because of the advent of generative AI (artificial intelligence). Most of these narratives about struggling college graduates can be traced back to an April 2025 New York Fed report on college graduate unemployment, and a subsequent report by consulting group Oxford Economics. Both sources note that the unemployment rate for recent college graduates in the U.S. now exceeds the overall unemployment rate. But a closer look at the data used in these reports suggests two things:
1. The timing of this trend doesn’t fit the AI-related explanations.
2. The trend is probably less about struggling college graduates, and more about a longer-term shift in labor force participation among young non-college workers.
Analysis
By replicating the New York Fed’s charts using Current Population Survey (CPS) microdata, we provide a more straightforward comparison of young workers ages 22 to 27, with and without a four-year college degree (see Figure 1). This is a more natural comparison, especially since recent trends in the “all workers” category might be confounded by the continued aging of the working-age population.
Figure 1: Unemployment rate for recent college graduates, young non-college workers and all workers
Note: Unemployment rate data is seasonally adjusted and shows a three-month average. Recent college graduates and young non-college workers are ages 22 to 27. Data is shown through June 2025.
Source: IPUMS-Current Population Survey
Still, it is clear that the gap in unemployment rates between recent college graduates and young non-college workers has narrowed recently. Looking at the difference in unemployment rates in Figure 2, the gap shrank from 3.8 percentage points in 2006, to 2.4 in 2019, then slightly further to 1.9 in the first half of 2025. At first glance, this suggests some erosion of the historic college graduate advantage in the labor market. But note that most of the closing of this gap happened prior to 2019, and the gap has been steady from 2023 to 2025, the period during which generative AI saw its largest leaps in capability and usage.
Figure 2: The difference between young college and non-college worker unemployment rates (in percentage points)
Note: Unemployment rate data is seasonally adjusted and shows annual averages. 2025 data could only be calculated using data through June.
Source: IPUMS-Current Population Survey
Additionally, this decreasing college graduate advantage in unemployment rates appears not to be driven by changes in employment. Instead, there seems to be a growing gap in labor force participation between college and non-college young people. Young people without college degrees have always had much lower labor force participation than college graduates, but between 2006 and 2025, this gap widened (see Figure 3). Participation among non-college young people fell 2.6 percentage points, from 78.5% to 75.9%. The rate for young college graduates also fell, but only by 0.8 percentage points from 86.3% to 85.5%. This divergence in labor force participation between young college graduates and non-college young people could account for all the convergence in the gap between these groups' unemployment rates over this period.
Figure 3: Labor force participation rates for recent college graduates and non-college young people
Note: Labor force participation rate data is seasonally adjusted and shows annual averages. Recent college graduates and young non-college workers are ages 22 to 27. 2025 data could only be calculated using data through June.
Source: IPUMS-Current Population Survey
A simpler way to see this is to look at the trend in the employment-to-population ratio, which is calculated independent of labor force participation. The employment-to-population ratio was 10.6 percentage points higher for recent college graduates compared to non-college young people in 2006, and is 10.7 points higher in the first half of 2025 (see Figure 4).
Figure 4: Employment-to-population ratio for recent college graduates and non-college young people
Note: Employment-to-population ratios are seasonally adjusted and show annual averages. Recent college graduates and young non-college workers are ages 22 to 27. 2025 data could only be calculated using data through June.
Source: IPUMS-Current Population Survey
Reasons for leaving the labor force
Between 2006 and 2025, there is a marked increase in both college and non-college young people citing school or training as a reason for not working (see Figure 5). There is also a decrease in those not working due to home or family responsibilities. These changes roughly offset each other. But for non-college young people, there has also been a substantial increase in the percent who are not working because of disability or illness, and those who give other reasons. Together, these are about the size of the overall change in this group’s labor force participation. It doesn’t appear that discouragement about job prospects is to blame for labor force participation trends of either group.
Figure 5: Contributions to change since 2006 in labor force participation for young workers
Note: Labor force participation rates are seasonally adjusted annual averages. Change is calculated since 2006. Recent college graduates and young non-college workers are ages 22 to 27. 2025 data could only be calculated using data through June.
Source: IPUMS-Current Population Survey
Conclusion
The historic unemployment rate advantage of young college graduates over non-college workers has indeed narrowed in the past two decades. Most of this decline occurred prior to growth in the use of generative AI, and one likely contributor to this decline is that labor force participation among young non-college people has declined over the past two decades. Overall employment rates show a stable employment advantage for young college graduates over non-college young people. The future impact of generative AI on labor markets is an important question, but recent claims about its effect on the well-established advantages of college-educated workers in the labor market appear to be premature.